Cyber-physical system based spatiotemporal distribution prediction method for electric vehicle charging load

Abstract

The application belongs to the technical field of electric vehicle charging load prediction, and particularly relates to a kind of electric vehicle charging load space-time distribution prediction method based on information physical fusion, method first generates the parking time of single electric vehicle, initial travel time, initial electric quantity, starting point position, end point position, obtains the shortest travel path of electric vehicle through shortest path planning, then calculates the maximum distance of electric vehicle based on the remaining electric quantity when electric vehicle generates charging demand, and further determines the final charging node, obtains the shortest charging path through shortest path planning, and then calculates the remaining battery power and queuing time when reaching the final charging node, and further determines the final charging node, start charging time and charging power of electric vehicle, and finally outputs the dynamic prediction result of electric vehicle charging load space-time distribution. The application can improve the prediction accuracy of charging load space-time distribution dynamic prediction result.

Description

Technical Field

[0001] This invention belongs to the field of electric vehicle charging load prediction technology, specifically relating to a method for predicting the spatiotemporal distribution of electric vehicle charging load based on cyber-physical fusion. Background Technology

[0002] Electric vehicles (EVs), as a promising green transportation tool, are an important means of tapping the carbon reduction potential of transportation and improving the electrification level of transportation. With the continuous growth in the number of EVs, huge charging demand will be generated, stimulating the comprehensive deployment and rapid growth of charging infrastructure. However, compared with the continuously growing market demand, the overall vehicle-to-charging-pile ratio still lags behind. Long charging times and poor charging experience are the main difficulties and challenges faced by existing EV charging technologies. Due to the lack of actual EV usage data in the initial stage, the construction of charging infrastructure lacks support and guidance, resulting in unreasonable planning of charging station locations and a mismatch between the construction of charging piles and actual demand. This leads to very high charging waiting rates in some areas and a large number of idle charging piles in remote areas. This contradiction has become a significant challenge for the industry. Utilizing computational intelligence to rationally and efficiently plan the layout and capacity of charging stations, adapting them to the actual spatiotemporal distribution of charging load demand, is an urgent solution to address the limited driving range of EVs, improve the charging experience, and increase the utilization rate of regional charging stations. This requires more accurate spatiotemporal prediction of EV charging load, providing an important basis for determining the location and capacity of EV charging stations.

[0003] Patent application number 202010013815.9 provides a method for predicting the spatiotemporal distribution of electric vehicle charging load based on a travel probability matrix. First, a probabilistic model of influencing factors is established. Then, based on the network topology of electric vehicles and their inter-city travel, an electric vehicle travel probability matrix is ​​constructed. Finally, the Monte Carlo method is used to predict the spatiotemporal distribution of electric vehicle charging load for a given day, based on the probabilistic model of influencing factors and the electric vehicle travel probability matrix. While this method can predict the electric vehicle charging load under spatiotemporal distribution, it does not consider the spatial perception of electric vehicle users regarding driving distance and congestion avoidance behavior, thus failing to meet the needs of electric vehicle users to quickly reach their destinations in actual driving, resulting in a relatively large prediction error. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned problems in the existing technology by providing a cyber-physical fusion-based method for predicting the spatiotemporal distribution of electric vehicle charging loads that meets the actual driving needs of electric vehicle users to quickly reach their destinations and has high prediction accuracy.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] This invention provides a method for predicting the spatiotemporal distribution of electric vehicle charging load based on cyber-physical fusion, the method comprising:

[0007] S1. Obtain the number of electric vehicles in use;

[0008] S2. Generate the parking time, initial travel time, initial battery level, starting position, and ending position of a single electric vehicle. Use the comprehensive road impedance model to perform shortest path planning to obtain the shortest travel path from the starting position to the ending position of the electric vehicle. Use the power consumption model combined with the shortest travel path to calculate the remaining battery power of the electric vehicle in real time. Determine in real time whether the remaining battery power meets the charging conditions. When the charging conditions are met, the electric vehicle generates a charging demand. Record the time when the electric vehicle generates a charging demand and proceed to S3.

[0009] S3. Calculate the maximum driving distance of the electric vehicle based on the remaining power when the electric vehicle generates a charging demand. Determine the set of optional charging nodes based on the maximum driving distance and the location where the electric vehicle generates a charging demand. Randomly select a charging node from the set of optional charging nodes as the final charging node and record the final charging node.

[0010] S4. Use the comprehensive road impedance model to perform shortest path planning to obtain the shortest charging path for the electric vehicle from the location where the charging demand is generated to the final charging node. Use the power consumption model in combination with the shortest charging path to calculate the remaining battery power of the electric vehicle when it arrives at the final charging node from the location where the charging demand is generated. Record the time when the electric vehicle arrives at the final charging node.

[0011] S5. Calculate the time required for the electric vehicle to queue for charging after arriving at the final charging node. Start charging after queuing. Select the charging mode based on the parking time of the electric vehicle and the remaining battery power when arriving at the final charging node. Record the start charging time and the charging power of the selected charging mode.

[0012] S6. Determine whether the recording of all final charging nodes, start charging time and charging power of electric vehicles has been completed. If not, return to step S2 to calculate the next electric vehicle. If yes, output the dynamic prediction result of the spatiotemporal distribution of electric vehicle charging load.

[0013] The method for obtaining the number of electric vehicles is as follows: select electric vehicle penetration rate, electric vehicle range, number of public charging piles in the region, GDP per capita, and government subsidies as input data to train a BP neural network to predict the number of electric vehicles.

[0014] The simulated annealing algorithm is used to optimize the weights of a backpropagation (BP) neural network. The specific steps are as follows:

[0015] S11. Set the initial temperature and cooling rate of the simulated annealing algorithm; generate the initial solution, i.e., randomly generate the initial weight set of the BP neural network;

[0016] S12. Generate a new solution based on the current solution, and calculate the fitness difference between the current solution and the new solution. ,like Then accept the new solution, if The Metropolis criterion then determines whether to accept the new solution; the Metropolis criterion states that if the randomly generated random number is less than the probability... Then accept the new solution, with a probability of , The current temperature must be used; otherwise, new solutions will not be accepted.

[0017] S13. Update the current solution; update the temperature according to the following formula: ,in , These represent the temperatures before and after the update, respectively. Indicates the cooling rate;

[0018] S14. Return to S12 for iterative optimization until the maximum number of iterations is reached. Select the current solution as the optimal weights to train the BP neural network.

[0019] The shortest path planning using the comprehensive road impedance model refers to obtaining the comprehensive road impedance of each road segment in the traffic network through the comprehensive road impedance model, and using the comprehensive road impedance of each road segment as the weight of each road segment to perform shortest path planning using the Floyd algorithm.

[0020] The comprehensive road impedance model includes:

[0021] ;

[0022] ;

[0023] ;

[0024] In the above formula, express Time Node To node The combined road impedance between the roads; Represents a node To node Travel time between roads; express At every moment Waiting time for traffic lights; express Road saturation at any time Time node To node Travel time between roads; Indicates road saturation hour Time Node To node Travel time between roads; express Road saturation at any time At the node Waiting time for traffic lights; express Road saturation at any time At the node Waiting time for traffic lights; This indicates the travel time required for an electric vehicle to pass through a road when there is zero traffic flow. , All are impact factors; For road saturation, This indicates that the road is clear. This indicates that traffic is moving slowly. Indicates road congestion. This indicates severe traffic congestion. , For road traffic flow, Road capacity refers to the number of vehicles that can pass smoothly through a road per unit of time. The signal cycle refers to the time required for a traffic light to complete one cycle of red-green-yellow. It is the ratio of the green light signal duration to the total signal light duration. This indicates the vehicle arrival rate, i.e. Time to reach node The number of electric vehicles.

[0025] The power consumption model includes:

[0026] ;

[0027] ;

[0028] ;

[0029] In the above formula, Remaining battery level; This is the initial charge level; For the battery capacity of electric vehicles; For nodes To node The length of the road between them; Energy consumption per unit mileage for electric vehicles; For electric vehicles at a driving speed of The driving distance is The ambient temperature is Air conditioner power consumption at that time; For electric vehicles, based on driving speed The electricity consumed while driving on the road; , Electric vehicles are respectively rated by driving speed. Electricity consumption generated when driving on main roads and secondary roads; , These are the air conditioning cooling and heating power, respectively. , These are the cold and hot limit temperatures, respectively. Indicates in The speed of the electric vehicle during the specified time period; Design vehicle speed for roads; Road capacity refers to the number of vehicles that can pass smoothly through a road per unit of time. In order to be in The number of vehicles traveling on the road during a given time period; It is a nonlinear function; , , All are adaptive parameters.

[0030] The following formula can be used to calculate the time required for an electric vehicle to queue for charging after arriving at the final charging node:

[0031] ;

[0032] ;

[0033] ;

[0034] ;

[0035] In the above formula, Waiting times for electric vehicles; This refers to the number of charging stations in the charging node. The number of electric vehicles that can be served by each charging station per unit of time; Service intensity for charging piles; The probability of receiving charging services for electric vehicles; Indicates the first One charging station; express factorial; express factorial; For average team leader; This represents the number of charging requests per hour to a charging station.

[0036] The specific method for selecting the charging mode based on the parking time of the electric vehicle and the remaining battery power when reaching the final charging node is as follows: calculate the time required for the electric vehicle to charge to the desired charging level in slow charging mode according to the following formula and determine whether it is greater than the parking time. If so, charge in fast charging mode; otherwise, charge in slow charging mode.

[0037] ;

[0038] In the above formula, The time required to charge an electric vehicle to the desired charging level in slow charging mode; The desired power consumption; The remaining battery power of the electric vehicle when it reaches the final charging point; For the battery capacity of electric vehicles; The charging power of electric vehicles in slow charging mode; The charging efficiency of electric vehicles.

[0039] The charging condition refers to the battery's remaining power reaching the charging threshold.

[0040] The expression for the spatiotemporal distribution of the electric vehicle charging load is:

[0041] ;

[0042] In the above formula, for Charging nodes The charging load at the location; for Charging nodes The charging power of the i-th electric vehicle connected at the selected charging mode.

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0044] 1. The present invention discloses a method for predicting the spatiotemporal distribution of electric vehicle charging load based on cyber-physical fusion. This method generates the parking duration, initial travel time, initial battery level, starting location, and ending location of a single electric vehicle. It uses a comprehensive road impedance model to perform shortest path planning to obtain the shortest travel path from the starting location to the ending location. It then uses a power consumption model combined with the shortest travel path to calculate the remaining battery power of the electric vehicle in real time, and determines in real time whether the remaining battery power meets the charging conditions. When the charging conditions are met, the electric vehicle generates a charging demand, and the time when the charging demand is generated is recorded. Subsequently, based on the remaining battery power at the time the charging demand is generated, the maximum drivable distance of the electric vehicle is calculated. Based on the maximum drivable distance and the location where the charging demand is generated, a set of optional charging nodes is determined. A charging node is randomly selected from the set of optional charging nodes as the final charging node, and the final charging node is recorded. The comprehensive road impedance model is used to perform shortest path planning to obtain the shortest charging path from the location where the charging demand is generated to the final charging node. Finally, the power consumption model is used to calculate the shortest charging path from the location where the charging demand is generated to the final charging node. The invention sets the remaining battery power upon arrival at the final charging node, records the time taken for the electric vehicle to reach the node, calculates the waiting time after arrival, and begins charging after queuing. Based on the vehicle's parking time and remaining battery power, a charging mode is selected, and the start charging time and charging power are recorded. After recording all final charging nodes, start charging times, and charging power, the dynamic prediction result of the spatiotemporal distribution of electric vehicle charging load is output. This design, on the one hand, performs shortest path planning for both the electric vehicle's journey to its destination and its journey to the charging node after generating a charging demand. This aligns with users' perception of the shortest travel distance and meets their need for quick arrival in actual driving, enabling more accurate determination of charging nodes and a more accurate description of the spatiotemporal distribution of electric vehicles. On the other hand, by calculating queuing time, it describes the impact of charging station congestion on the spatiotemporal distribution of electric vehicles, ultimately improving the prediction accuracy of the dynamic prediction result of the spatiotemporal distribution of electric vehicle charging load. Therefore, this invention meets the needs of electric vehicle users for quick arrival in actual driving and improves prediction accuracy.

[0045] 2. The electric vehicle charging load spatiotemporal distribution prediction method based on cyber-physical fusion described in this invention combines model-driven and data-driven approaches. Specifically, it first uses a BP neural network optimized by simulated annealing to predict the number of electric vehicles, and then dynamically predicts the spatiotemporal distribution of charging load based on the predicted number of electric vehicles. This method is suitable for application in areas where large-scale electric vehicle access is planned. Therefore, this invention is suitable for application in areas where large-scale electric vehicle access is planned.

[0046] 3. The electric vehicle charging load spatiotemporal distribution prediction method based on cyber-physical fusion described in this invention, in its constructed comprehensive road impedance model, not only integrates road segment impedance and node impedance caused by generation time delays under traffic light control, but also further divides the impedance according to road saturation. This better aligns with users' spatial perception of congested roads ahead in actual driving behavior, thereby making the shortest path planning results more accurate and further improving the accuracy of the charging load spatiotemporal distribution prediction results. Therefore, this invention conforms to users' spatial perception of congested roads ahead in actual driving behavior and can further improve prediction accuracy. Attached Figure Description

[0047] Figure 1 This is a flowchart of the spatiotemporal distribution prediction method for charging load described in this invention. Detailed Implementation

[0048] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0049] Example:

[0050] See Figure 1 A method for predicting the spatiotemporal distribution of electric vehicle charging load based on cyber-physical fusion is proposed, which proceeds in the following steps:

[0051] S1. Obtain the number of electric vehicles in use. The method is as follows: select electric vehicle penetration rate, electric vehicle range, number of public charging piles in the region, GDP per capita, and government subsidies as input data to train a BP neural network to predict the number of electric vehicles in use. Since the initial weights of the BP neural network have not been optimized, it is prone to getting trapped in local minima and has a slow convergence speed. Therefore, simulated annealing algorithm is used to optimize the weights of the BP neural network. The specific steps are as follows:

[0052] S11. Set the initial temperature and cooling rate of the simulated annealing algorithm. The initial temperature is used to determine the randomness of the search, and the cooling rate is used to decay the temperature of the next round according to the current temperature. Generally, the cooling rate is set to 0.95. Generate the initial solution, that is, randomly generate the initial weight set of the BP neural network.

[0053] S12. Generate a new solution based on the current solution;

[0054] S13. Calculate the fitness difference between the current solution and the new solution. ,like Then accept the new solution, if The Metropolis criterion then determines whether to accept the new solution; the Metropolis criterion states that if the randomly generated random number is less than the probability... Then accept the new solution, with a probability of , The current temperature must be used; otherwise, new solutions will not be accepted.

[0055] S14. Update the current solution;

[0056] S15. Treat S12-S14 as one iterative optimization process. After each iteration, update the temperature according to the following formula: ,in , These represent the temperatures before and after the update, respectively. Indicates the cooling rate; the updated temperature is used as the current temperature.

[0057] S16. Return to S12 for the next iteration optimization, until the maximum number of iterations is reached;

[0058] S17. Select the current solution as the optimal weights to train the BP neural network;

[0059] S2. Read the type of a single electric vehicle and generate its basic parameters based on that type. These parameters include parking duration, initial travel time, battery capacity, and initial charge level. The initial travel time refers to the time the electric vehicle first departs from its starting point within a day. For example, electric vehicles include three types: taxis, private cars, and urban utility vehicles. Different types of electric vehicles have different parking durations, initial travel times, battery capacities, and initial charge levels, which can be obtained by statistically analyzing the probability distribution of the basic parameters corresponding to each electric vehicle type. The destination position of the electric vehicle is then generated based on its starting point and the travel probability matrix.

[0060] The shortest path is obtained for the electric vehicle from the starting point to the ending point by using the comprehensive road impedance model for shortest path planning; the remaining battery power of the electric vehicle is calculated in real time by using the power consumption model in combination with the shortest travel path, and it is determined in real time whether the remaining battery power has reached the charging condition. The charging condition refers to the remaining battery power reaching the charging threshold, which is generally set to 20% of the battery capacity; when the charging threshold is reached, the electric vehicle generates a charging demand, the time when the electric vehicle generates a charging demand is recorded, and the process proceeds to S3.

[0061] Shortest path planning using a comprehensive road impedance model refers to obtaining the comprehensive road impedance of each road segment in the traffic network through the comprehensive road impedance model, and using the comprehensive road impedance of each road segment as the weight to perform shortest path planning using the Floyd algorithm. Since electric vehicles are affected not only by the road's inherent resistance but also by traffic light control at road nodes when passing through roads, resulting in greater time delays under traffic light signal control, this invention integrates road impedance and node impedance caused by time delays under traffic light control when designing the comprehensive road impedance model. The comprehensive road impedance model includes:

[0062] ;

[0063] ;

[0064] ;

[0065] In the above formula, express Time Node To node The combined road impedance between the roads; Represents a node To node Travel time between roads; express At every moment Waiting time for traffic lights; express Road saturation at any time Time node To node Travel time between roads; Indicates road saturation hour Time Node To node Travel time between roads; express Road saturation at any time At the node Waiting time for traffic lights; express Road saturation at any time At the node Waiting time for traffic lights; This indicates the travel time required for an electric vehicle to pass through a road when there is zero traffic flow. , All are impact factors; For road saturation, This indicates that the road is clear. This indicates that traffic is moving slowly. Indicates road congestion. This indicates severe traffic congestion. , For road traffic flow, Road capacity refers to the number of vehicles that can pass smoothly through a road per unit of time. The signal cycle refers to the time required for a traffic light to complete one cycle of red-green-yellow. It is the ratio of the green light signal duration to the total signal light duration. This indicates the vehicle arrival rate, i.e. Time to reach node The number of electric vehicles;

[0066] This invention considers the impact of road capacity and ambient temperature on power consumption when designing the power consumption model, thereby more accurately describing user driving habits under complex road conditions; the power consumption model includes:

[0067] ;

[0068] ;

[0069] ;

[0070] In the above formula, Remaining battery level; This is the initial charge level; For the battery capacity of electric vehicles; For nodes To node The length of the road between them; Energy consumption per unit mileage for electric vehicles; For electric vehicles at a driving speed of The driving distance is The ambient temperature is Air conditioner power consumption at that time; For electric vehicles, based on driving speed The electricity consumed while driving on the road; , Electric vehicles are respectively rated by driving speed. Electricity consumption generated when driving on main roads and secondary roads; , These are the air conditioning cooling and heating power, respectively. , These are the cold and hot limit temperatures, respectively. Indicates in The speed of the electric vehicle during the specified time period; Design vehicle speed for roads; Road capacity refers to the number of vehicles that can pass smoothly through a road per unit of time. In order to be in The number of vehicles traveling on the road during a given time period; It is a nonlinear function; , , All are adaptive parameters;

[0071] S3. Calculate the maximum driving distance of the electric vehicle based on the remaining power when the electric vehicle generates a charging demand. Determine the set of optional charging nodes based on the maximum driving distance and the location of the electric vehicle when it generates a charging demand. Randomly select a charging node from the set of optional charging nodes as the final charging node and record the final charging node.

[0072] S4. Use the comprehensive road impedance model to perform shortest path planning to obtain the shortest charging path for the electric vehicle from the location where the charging demand is generated to the final charging node. Use the power consumption model in combination with the shortest charging path to calculate the remaining battery power of the electric vehicle when it arrives at the final charging node from the location where the charging demand is generated. Record the time when the electric vehicle arrives at the final charging node.

[0073] S5. To accurately describe the congestion situation at charging stations, the occupancy rate of charging piles at the final charging node is statistically calculated for each time period. If the occupancy rate of charging piles is less than or equal to 1 when an electric vehicle arrives at the final charging node, the electric vehicle can charge directly without waiting. The queuing time at this time is... If the charging station occupancy rate is 0, and the occupancy rate is greater than 1, then the time required for an electric vehicle to queue for charging after arriving at the final charging node is calculated using the following formula:

[0074] ;

[0075] ;

[0076] ;

[0077] ;

[0078] In the above formula, The time required for electric vehicles to wait in line to charge; This refers to the number of charging stations in the charging node. The number of electric vehicles that can be served by each charging station per unit of time; Service intensity for charging piles; The probability of receiving charging services for electric vehicles; Indicates the first One charging station; express factorial; express factorial; For average team leader; The number of charging demands per hour that go to a charging station for service;

[0079] After queuing, charging begins. First, the charging mode is selected based on the parking time of the electric vehicle and the remaining battery power when it reaches the final charging node. Then, the time required for the electric vehicle to charge to the desired charging level under the selected charging mode is calculated. The charging start time, the charging power and charging time under the selected charging mode are recorded. The charging time of the electric vehicle must not exceed the parking time.

[0080] The charging mode selection based on the parking time of the electric vehicle and the remaining battery power when reaching the final charging node is as follows: calculate the time required for the electric vehicle to charge to the desired charging level in slow charging mode according to the following formula and determine whether it is greater than the parking time. If so, charge in fast charging mode; otherwise, charge in slow charging mode.

[0081] ;

[0082] In the above formula, The time required to charge an electric vehicle to the desired charging level in slow charging mode; The desired power consumption; The remaining battery power of the electric vehicle when it reaches the final charging point; For the battery capacity of electric vehicles; The charging power of electric vehicles in slow charging mode; For electric vehicle charging efficiency;

[0083] The formula for calculating the time required for an electric vehicle to reach the desired charging level using slow charging mode is:

[0084] ;

[0085] In the above formula, The time required to charge an electric vehicle to the desired charging level in fast charging mode; The charging power of electric vehicles in fast charging mode;

[0086] S6. Determine whether the recording of all final charging nodes, start charging times, charging power, and charging duration for all electric vehicles has been completed. If not, return to step S2 to calculate for the next electric vehicle. If the calculation of the charging power for all electric vehicles has been completed, output the dynamic prediction result of the spatiotemporal distribution of the electric vehicle charging load. The expression for the spatiotemporal distribution of the electric vehicle charging load is:

[0087] ;

[0088] In the above formula, for Charging nodes The charging load at the location; for Charging nodes The charging power of the i-th electric vehicle connected at the selected charging mode.

Claims

1. A cyber-physical system-based method for predicting the spatiotemporal distribution of electric vehicle charging loads, characterized in that the method comprises the following steps: S1. Obtain the number of electric vehicles; S2. Obtain the parking duration, initial travel time, initial battery level, starting location, and ending location of each electric vehicle, and use a comprehensive road impedance model to plan the shortest path from the starting location to the ending location, and use a battery consumption model to calculate the remaining battery level in real time, and determine whether the remaining battery level meets the charging condition, and if so, record the time when the electric vehicle generates a charging demand, and proceed to S3; S3. Calculate the maximum distance that the electric vehicle can travel based on the remaining battery level when the electric vehicle generates a charging demand, and determine the set of optional charging nodes based on the maximum distance and the location where the electric vehicle generates a charging demand, and randomly select one charging node from the set of optional charging nodes as the final charging node, and record the final charging node; S4. Use the comprehensive road impedance model to plan the shortest charging path from the location where the electric vehicle generates a charging demand to the final charging node, and use the battery consumption model to calculate the remaining battery level when the electric vehicle arrives at the final charging node, and record the time when the electric vehicle arrives at the final charging node; S5. Calculate the time required for the electric vehicle to queue for charging after arriving at the final charging node, and start charging after the queue is completed, and select a charging mode based on the parking duration and the remaining battery level when the electric vehicle arrives at the final charging node, and record the start time of charging and the charging power under the selected charging mode; S6. Determine whether the recording of the final charging node, the start time of charging, and the charging power of all electric vehicles has been completed, and if not, return to step S2 to calculate the next electric vehicle, and if so, output the dynamic prediction result of the spatiotemporal distribution of electric vehicle charging loads. The comprehensive road impedance model comprises:

2. The cyber-physical system-based method for predicting the spatiotemporal distribution of electric vehicle charging loads according to claim 1, characterized in that the method for obtaining the number of electric vehicles comprises the following steps: ； ； ； In the above formula, represents the time node to node between the road; represents the node to node between the road between the travel time; represents the time node waiting for the traffic light; represents the time road saturation time node to node between the road between the travel time; represents the road saturation time the time node to node between the road between the travel time; represents the time road saturation the time node waiting for the traffic light; represents the time road saturation the time node waiting for the traffic light; represents the travel time required for an electric vehicle to pass through the road when the road has zero traffic flow; , are all impact factors; is the road saturation, represents the road smooth, represents the road slow, represents the road congestion, represents the road serious congestion, , is the road traffic flow, is the road traffic capacity, that is, the number of vehicles that can pass through the road smoothly per unit time; is the signal cycle, which refers to the time required for a traffic signal to go through a red-green-yellow cycle; is the ratio of the length of the green light signal to the length of all signal lights; represents the vehicle arrival rate, that is the number of electric vehicles arriving at the node at the time. Select the electric vehicle penetration rate, electric vehicle range, number of public charging stations in the region, per capita GDP, and government subsidies as input data to train a BP neural network, and predict the number of electric vehicles.

3. The cyber-physical system-based method for predicting the spatiotemporal distribution of electric vehicle charging loads according to claim 2, characterized in that the weights of the BP neural network are optimized using a simulated annealing algorithm, and the specific steps are as follows: S11. Set the initial temperature and cooling rate of the simulated annealing algorithm, and generate an initial solution, i.e., randomly generate an initial weight set of the BP neural network; S13. Update the current solution; S14. Return to S12 for iterative optimization until the maximum number of iterations is reached, and select the current solution as the optimal weight to train the BP neural network. S12, generating a new solution on the basis of the current solution, calculating the fitness difference between the current solution and the new solution , if then accepting the new solution, if then deciding whether to accept the new solution according to the Metropolis criterion; the Metropolis criterion is that if a random number generated randomly is less than a probability then accepting the new solution, the probability , is the current temperature, otherwise not accepting the new solution; ​ The temperature is updated according to the following equation: wherein , Toldand Tnewdenote the temperature before and after the update, respectively, denotes the cooling rate; ​ 4. The cyber-physical system based spatio-temporal distribution prediction method of electric vehicle charging load according to any one of claims 1-3, characterized in that: The shortest path planning by using the comprehensive road impedance model refers to obtaining the comprehensive road impedance of each road section in the traffic network by using the comprehensive road impedance model, and using the comprehensive road impedance of each road section as the weight of each road section to perform the shortest path planning by using the Floyd algorithm.

5. The cyber-physical system based spatio-temporal distribution prediction method of electric vehicle charging load according to any one of claims 1-3, characterized in that: The power consumption model comprises: ； ； ； In the above formula, is the remaining power; is the initial power; is the battery capacity of the electric vehicle; is the node is the node is the road length between nodes is the unit energy consumption per mile of the electric vehicle; is the air conditioning power consumption of the electric vehicle when the driving speed is , the driving distance is , and the ambient temperature is ; is the power consumption of the electric vehicle driving on the road at a driving speed of ; , are the power consumptions of the electric vehicle driving on the main road and secondary road at a driving speed of ; , are the air conditioning refrigeration and heating power; , are the cold and hot limit temperatures; represents the driving speed of the electric vehicle at the time period; is the road design speed; is the road capacity, i.e., the number of vehicles that can pass smoothly on the road per unit time; is the number of vehicles driving on the road at the time period; is a nonlinear function; , , are all adaptive parameters.

6. The cyber-physical system based spatio-temporal distribution prediction method of electric vehicle charging load according to claim 5, characterized in that: The time length required for the electric vehicle to queue for charging after reaching the final charging node is calculated according to the following formula: ； ； ； ； In the above formula, is the queue length of electric vehicles; is the number of charging piles in the charging node; is the number of electric vehicles that can be served per unit time for each charging pile; is the service intensity of the charging pile; is the probability of electric vehicles accepting charging service; represents the charging pile; represents factorial; represents factorial; is the average queue length; is the charging demand per hour to go to the charging station for service.

7. The cyber-physical system based spatio-temporal distribution prediction method of electric vehicle charging load according to claim 6, characterized in that: The selection of the charging mode based on the parking time length of the electric vehicle and the remaining battery power when reaching the final charging node is specifically that the time length required for the electric vehicle to charge to the expected charging level in the slow charging mode is calculated according to the following formula, and it is determined whether the time length is greater than the parking time length, if yes, the electric vehicle is charged in the fast charging mode, otherwise, the electric vehicle is charged in the slow charging mode. ； In the above formulae, is the time duration required for the electric vehicle to be charged to the desired charge level in slow charging mode; is the desired electric quantity; is the remaining electric quantity of the electric vehicle when it reaches the final charging node; is the battery capacity of the electric vehicle; is the charging power of the electric vehicle in slow charging mode; is the charging efficiency of the electric vehicle.

8. The cyber-physical system based spatio-temporal distribution prediction method of electric vehicle charging load according to any one of claims 1-3, characterized in that: The charging condition refers to that the remaining battery power reaches a charging threshold.

9. The cyber-physical system based spatio-temporal distribution prediction method of electric vehicle charging load according to any one of claims 1-3, characterized in that: The expression of the spatio-temporal distribution of the electric vehicle charging load is: ； In the above formula, for Charging nodes The charging load at the location; for Charging nodes The charging power of the i-th electric vehicle connected at the selected charging mode.

Images